Document 11163189

advertisement
working paper
department
of economics
-''^'''^^'^^^'
THE POTENTIAL FOR REDUCING CARBON EMISSIONS
FROM INCREASED EFFICIENCY:
A GENERAL EQUILIBRIUM METHODOLOGY
Charles R. Blitzer
Richard S Eckaus
Supriya Lahiri
Alexander Meeraus
.
No.
559
May 1990
massachusetts
institute of
_
technology
^
50 memorial drive
Cambridge, mass. 02139
THE POTENTIAL FOR REDUCING CARBON EMISSIONS
FROM INCREASED EFFICIENCY:
GENERAL
EQUILIBRIUM METHODOLOGY
A
Charles R. Blitzer
Richard S Eckaus
Supriya Lahiri
Alexander Meeraus
.
No.
559
May 1990
THE POTENTIAL FOR REDUCING CARBON EMISSIONS
FROM INCREASED EFFICIENCY:
A GENERAL EQUILIBRIUM METHODOLOGY
Charles R. Blitzer
World Bank
Richard S. Eckaus
M.I.T.
Supriya Lahiri
University of Lowell
Alexander Meeraus
Corporation
Development
GAMS
May 1990
Center for Energy Policy Research
M.I.T.
The views expressed here are the authors' responsibility and
do not necessarily reflect the views of the Center for Energy
Policy Research, the University of Lowell, the Massachusetts Institute of Technology or the World Bank.
The project in which
the original model was developed was sponsored by an MIT subcontract with Meta Systems of Cambridge, Mass., contracting with
the Organization for Energy Policy, Government of Egypt and
financed by the U.S. Agency for International Development.
Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium IVIember Libraries
http://www.archive.org/details/potentialforreduOOblit
1
May 18, 1990
THE POTENTIAL FOR REDUCING CARBON EMISSIONS
FROM INCREASED EFFICIENCY:
A GENERAL EQUILIBRIUM METHODOLOGY
I.
Introduction
One approach to reduction of carbon emissions that has often
been suggested is through an increase in the efficiency of use of
hydrocarbon fuels, i.e., achieving more output for the same
amount or lesser fuel input.
It seems like an obviously
desirable prescription, particularly if it can be done with just
a little improvement in technique and tuning.
Something for
nothing is always a good trade.
The
argument has been generalized, however, to situations
in which better fuel efficiency requires more of other inputs,
including, usually, capital.
But,
becomes less obviously a good one.
tion go up?
in this case, the prescription
What if the costs of produc-
What if production of the capital and/or the other
inputs is, itself, relatively fuel intensive?
As is often the
case, a closer look reveals complexities that were not so obvious
at first glance.
This paper presents a methodology for analyzing the potential for reduction in carbon emissions through increased fuel ef-
ficiency and provides an illustration of the method.-'-
It is in-
tended as a demonstration of technique rather than a substantive
2
contribution, although it may help illustrate the general structure of possibilities.
The methodology employed is
a
multisectoral
,
intertemporal,
programming model embodying significant non-linearities in proThe demonstration uses a model that was
duction and consumption.
first constructed to analyze energy policy in Egypt.
It is,
therefore, a special case, representing a developing country that
utilizes only two hydrocarbon fuels, natural gas and petroleum,
with only a negligibly small amount of imported coal.
The results should be considered only as illustrative for
other reasons as well.
First, while the basic parameters were
estimated from existing data for the Egyptian economy, the numbers used for the alternative, more fuel efficient technologies
are point estimates made by the authors.
It adds only a modest
amount of credence to them to say that our conjectures were based
on our reading of the available literature.^
Second,
it is assumed that the constraints on carbon emis-
sions, that are used to illustrate the potential, can be effec-
tively enforced.
Conceivably that could be done by direct
regulation or through the use of taxes as indicated by the shadow
prices generated in the model's solution.
Third, sources of greenhouse gases other than due to the use
of hydrocarbons as fuels,
as well as sources of potential
amelioration of carbon emissions, are simply ignored.
Fourth, alternative energy technologies, not based on
hydrocarbon fuels, are not considered.
3
While making a point of warning of the limitations of the
demonstration, the authors believe that, at this relatively early
stage of economic/environmental modeling, it is useful to present
the method.
II.
Fuel efficiency and carbon emissions
There are a number of sources of the "greenhouse effect" of
which carbon emissions are only one.
fuels,
in turn,
emissions.
The use of hydrocarbon
is only one among a number of sources of carbon
It is, however, one of the most important sources,
and has, therefore, been the focus of a good deal of attention.
Analysis of the economic consequences of both greenhouse effects and policies undertaken to prevent those effects are both
at an early stage.
Nonetheless it is transparently obvious that
constraints on the use of fuels to reduce carbon emissions would,
all other things equal, depress economic activity.-^
To
ameliorate the consequences of assuming, "all other things
equal," attention has been given to the potential for reduction
in greenhouse gases through an increase in hydrocarbon fuel ef-
ficiency, i.e., the output/fuel input ratio.
Whether that would
offset the higher costs of substitute energy sources depends, of
course, on the costs of obtaining higher fuel efficiency.
There have been a number of suggestive articles, which em-
body what might, at best, be described as
approach," to this problem.
a,
"partial equilibrium
The consequences of an improvement
in hydrocarbon fuel efficiency are treated as if the only effects
are the direct ones of a reduction in fuel requirements for any
particular level of output.
4
This type of problem requires
a
"general equilibrium" analy-
sis that takes into account both direct and indirect effects.
The latter include:
(1)
the necessary changes in other input re-
quirements, including capital requirements, in the process of
economic growth and
(2)
changes in prices and the induced changes
in composition of consumption and investment.
Improvements in fuel efficiency might come about in several
ways:
(1)
through changes in engineering and maintenance prac-
tices that require adjustments that are virtually costless;
(2)
through increased investment in new plant and equipment that uses
fuel more efficiently;
(3)
through technological changes that
permit lower fuel input/output ratios with the same or lower
amounts of capital and other inputs,.
It is reasonable to be skeptical with respect to the poten-
tial for improvements of the first type: costless increases in
fuel efficiency.
After all, if engineers and managers are profit
maximizers, as conventionally assumed, it would, in general, be
expected that any known adjustment that would reduce costs should
have been exploited.
If it has not been, the logic goes, the ad-
justment is not worth the effort.
On the other hand, there is a substantial literature that
attests to the pervasive differences between "best practice"
techniques and those that are in general use in production.
Har-
vey Leibenstein brought this to the attention of the economics
profession in a series of articles on, "X-inef f iciency
,
"
so-
called because it is not consistent with the conventional assumptions and methods of thoroughgoing profit maximization.^
'
5
Leibenstein associated such inefficiency with bureaucratic
obstacles, inertia and a failure to respond to small incentives.
In more contemporary language X-inef f iciency might be "explained"
in terms of information costs, transactions costs, and frictional
adjustment costs, giving a broad definition to the terms so that
they includes the effort necessary to change established patterns.
not,
In this sense, however, a reduction in X-inef f iciency is
in fact, costless, but requires,
ments that result in at least
a
at the least, rearrange-
temporary reduction in output, if
not an increase in other inputs.
The potential for increasing hydrocarbon fuel efficiency
through increased investment in plant and equipment deserves
careful investigation.
It should not be expected to be true, and
is patently not the case, that even the most modern technologies
are as fuel efficient as possible.
A rational manager should
maximize profits, which need not necessarily be the same objective as minimizing all costs and certainly not necessarily the
same objective as minimizing fuel costs.
However, as fuel costs
increase, or constraints are imposed on carbon emissions through
taxes or regulatory procedures, a shift to more fuel efficient
technologies should be expected.
There
ficiency.
a-re
many uncritical enthusiasts for increased fuel ef-
Yet the simple maxim, "when it pays, it will happen,"
is a good guideline.
It need not express complacency, but rather
may indicate the need for tax policy when markets do not fully
reflect any externalities associated with fuel usage.
6
Even so, there are good reasons to think that the maxim
might not have universal applicability.
The X-inef f iciency type
of argument has already been mentioned.
In addition,
only if
there are perfect capital markets will every project that can
meet the going risk and return criteria be undertaken.
The exis-
tence of credit rationing and other capital market imperfections
will prevent that from happening.
The potential gains in fuel efficiency from technological
changes have also been mentioned prominently as a source of
reductions in carbon emissions.
in making policies on this basis.
However, we should be cautious
We can wish for anything, but
that will not make it so.
III. An economy-wide and intertemporal environmental model
with alternative fuel efficiency possibilities
The model to be presented below has been used before by the
authors to analyze the effects on economic growth of constraints
on carbon emissions.
ing model and, thus,
It is a multisector,
intertemporal optimiz-
is in the same spirit as the approach by
Manne and Richels (1989).
However, as a result of its restricted
focus on a single country, it is more disaggregate and elaborate
in a number of respects than would have been warranted in Manne
and Richel's environmental modeling research.
The basic structure of the model is well-known from previous
work by the authors and many others.
The complete mathematical
structure of the model is presented in an appendix and only those
features that are particularly important for its present applica-
7
tion will be described here.
The model was originally con-
structed for the analysis of energy policy in Egypt.
It was
adapted to the analysis of environmental issues since it is rela-
tively detailed with respect to the sources and uses of energy,
which, as noted above, is one of the primary sources of environ-
mental offense.
For many purposes of environmental analysis, a country based
analysis is the correct one.
With some exceptions, as, for exam-
ple, the Montreal agreement on the control of fluorocarbon emis-
sions and regulation of the quality of the Rhine river, environ-
mental policies are now national, rather than international.
Economic policies, with only a few exceptions, are, likewise, national rather than international
There are, moreover, differences among countries in their
economic structures.
These are important in the analysis of the
effects of emissions constraints.
In particular, the relative
importance of the different productive sectors in developing
countries is changing relatively rapidly.
of these countries,
For example, in most
it should be expected that the weight of ag-
riculture in the economy will decline relative to industry and
that both production and consumption will become more energy intensive.
Nonetheless, some apologies are required for respecting national boundaries that are, for environmental purposes, often
quite artificial.
First, the local effects of some kinds of en-
vironmental pollution are the most important and averaging over a
8
larger area is misleading.
Second, transnational effects may be
not be confined only to border areas.
With these apologies, a
national model will be presented, but also with the belief that
the methodology is generalizable and extendable.
The model used here has a 25 year time horizon, divided into
five periods of five years each.
This somewhat artificial pacing
makes it possible to avoid a more detailed formulation of year-
by-year interactions and dynamic processes while still generating
a close
temporal approximation of growth conditions.
Results are
reported for five, evenly-spaced years.
The economy is divided into ten sectors, six of which are
non-energy sectors: agriculture, manufacturing, construction,
transportation, services and non-competing imports.
There are
four energy sectors: crude oil, natural gas, petroleum products
and electricity.
The economic variables determined by the model are investment, capital capacity and production by each sector, household
consumption by sector, energy demand and supply, imports and exports and relative prices, all calculated for each of five evenly
spaced, periods that are,
in turn,
five years apart.
As noted, the model focuses only on the generation of carbon
emissions due to fuel use, although the methods are adaptable to
other types of emissions associated with the use of any input or
to the output of particular goods with specific technologies.
The carbon emissions are calculated for each sector, as well as
in total,
for each period.
9
As an optimizing model,
it maximizes an objective or welfare
function which is the discounted sum of aggregate consumer utility over the model's horizon.
The utility of the representative
consumer in each time period is a weighted logarithmic sum over
all goods of the difference between its consumption of each type
of good and a parametrically fixed consumption level.
Individual
utility is multiplied by the projected population to obtain aggregate utility.
This formulation is identical to simulating the
market behavior of a representative consumer modeled as a linear
expenditure system.
It should be noted,
in the present context,
that environmental conditions do not enter the consumer's utility
function directly.
However, the consumer's choice of goods in
the consumption basket will depend on relative prices and income
levels, which are determined within the model, and those can be
expected to be affected by environmental policies.
The usual material balance constraints, which require that
the aggregate uses of output can be no greater than the aggregate
availabilities, apply in each period.
Availabilities depend on
domestic production and imports, where the latter is feasible.
One of the most significant features of the model for the
purposes of assessing the environmental impacts of economic ac-
tivity is that, in general, production of each good can be
carried out by alternative technologies, or, "activities," with
different input patterns.
The total output of each sector is the
sum of the production from each of the technologies.
Thus, there
is the possibility of substitution among inputs in production
10
processes.
The substitution is endogenously determined, in
response to the relative prices of inputs and outputs, which are
also determined endogenously.
This is important for the analysis
of environmental policies that either directly or indirectly af-
fect the cost of inputs.
The alternative requirements for production in each sector
are, with one exception,
specified exogenously, as if they were
taken from engineering specifications.
The exception is in the
demand for fuels, where, in effect, the BTU requirements per unit
of output are specified, but the requirements can be met by using
Here, again, the choice will be
either natural gas or petroleum.
made endogenously, depending on relative prices and any con-
straints that affect those prices.
Only three primary energy sources are distinguished:
hydropower, crude oil and natural gas.^
constrained by availability.
Production of each is
Crude oil is produced from
petroleum reserves and the creation and use of these and of natural gas reserves is modeled to reflect the fact that the level of
reserves is a function of the rate as well as the quantity of use
of the resources and outputs to producers and consumers.
For the present purposes, increases in fuel efficiency are
introduced in two alternative ways, as indicated briefly above.
First,
in a set of calculations,
increases in fuel efficiency are
incorporated in a costless manner simply by stipulating reduced
fuel input requirements.
Then increases in fuel efficiency are
associated with increases in capital input requirements.
11
Production also requires labor inputs, whose unit require-
ments are also specified exogenously, but differently, for each
technology or activity in each sector.
There is an overall con-
straint on labor availability and, separately, a labor constraint
in the agricultural sector intended to reflect limited rural-
urban labor mobility and the tightness of the rural labor market
over the past decade or so.
As is customary in such models, and different from the
Jorgenson/Wilcoxen and Nordhaus models, capital is specific to
each sector and, here, it is specific as well to the particular
technology that it embodies.
Capital formation in each period in
each sector requires that investment be undertaken in the pre-
vious period.
Depreciation rates are specified exogenously for
the capital stock used by each technology in each period.
Foreign trade is confined to the tradable goods sectors: agriculture, manufacturing, transportation, other services, crude
oil and petroleum products.
As an approximate way of recognizing
limited flexibility in the response of exports and imports to
changes in relative prices, the rate of change of each of these
is constrained, although within wide bounds.
The overall balance of payments constraint, that limits im-
ports to what can be paid for from exports and foreign exchange
resources, must also be met.
Foreign borrowing is allowed,
within moving upper bounds.
The problems of establishing initial and terminal conditions
in a model of this sort are well-known.
Here they are, largely.
12
finessed.
The sectoral levels of investment in the initial peri-
od are constrained not to be greater than those actually achieved
The sectoral levels of investment in the terminal peri-
in 1987.
od are determined by the condition that they be adequate to
sustain an exogenously specified rate of growth of output in the
sector in the post terminal period.
The terminal conditions
create some anomalies in the final periods of the model's time
Since that horizon is relatively long, these have only
horizon.
modest effects on the intermediate years.
With this description of the basic model in place it is possible now to turn to the features that deal with carbon emissions, which can,
in fact, be described quickly.
The quantity of
carbon, V, that is generated by the use of a particular fuel,
in a technology, k,
^ikit*
in a particular sector,
j,
in period, t,
i,
is
^° ^^® total amount of carbon generated by the use of a
particular fuel in the sector is obtained by summing over all
technologies
Vijt =
r
Vikjt
•
k
The total amount of carbon generated by the use of the particular
fuel in all sectors is:
V it
=
2
^ijt
2. ^ij^
J
The generation of carbon is related to the use of the particular
.
fuel in the sector by a coefficient,
vj^j^-;^.
I.e.,
^kijt = ^kijt ^kjt'
where the
Vj^j^'s
are understood to refer only to the fuel inputs.
The simple relationships are the conventional ones used in
projecting the generation of environmental agents.
Now, however.
13
that generation is a matter of endogenous determination in a com-
plete model.
So calculation of the generation of environmental
agents is completely consistent with the calculation of the other
features of the model, including its growth path.
The economic effects of carbon emissions constraints depends
on the manner in which they are imposed.
In a previous paper the
authors investigated the relative effects of global as compared
to sectoral emissions constraints.^
It was not surprising to
find that the latter had more depressing economic effects than
the former.
Attention will, therefore, be focused here on global
or aggregate carbon emissions constraints.
These take the form:
Z.Vit^Vt
i
This type of restriction can be used to reflect the idea of "bubble" regulation.
It is, essentially, regulating the total output
of an environmental agent by a complex of industries so as to
permit the individual industries to choose, themselves, the most
efficient means of meeting the overall target.
This type of restriction will be applied with greater or
lesser severity in various periods, in conjunction with different
rates of fuel efficiency to investigate the trade-offs between
reduction in the generation of carbon emissions and overall economic performance.
IV.
Perspectives on the model
In this model, all adjustments are optimal,
maximization of the objective function.
in terms of the
Moreover, they are made
with perfect foresight over the model's time horizon.
The im-
14
plicit assumption is that agents in the economy act efficiently
to maximize their welfare with perfect foresight.
A single solu-
tion of the model provides, therefore, what must be regarded as
an optimistic projection of what can be achieved in terms of the
maximand, given the endowments, the opportunities and constraints
that are represented in its framework.
While it is correct to
question the reality of such optimism, the approach does meet the
question often raised as to whether projected adjustments to
policy are the most efficient ones.
In any case, a particular solution to the model is of less
interest than the comparisons among solutions, which provide insights into problems and opportunities.
This is particularly
true when, as in the present application, the data represent only
rough approximations.
In the applications reported on here, the
comparison will be between economic outcomes with and without
carbon emission controls and with and without projected improve-
ments in fuel efficiency.
In all cases the solutions are dynami-
cally efficient with respect to the objective function.
In the comparisons to be made,
it is less clear that the
results with respect to the effects of emission constraints
should be interpreted as, "optimistic," since the basis for the
comparison is also always an optimal result.
Even without actually solving the model we know what the effects of emissions constraints must be.
If the constraints are
binding, and it is expected that they will be, economic per-
formance measured in terms of the objective function and the re-
15
lated output and income levels will suffer.
Only on the assump-
tion that there are costless ways of adjusting to the constraints
could the results be different.
It is plausible for advanced countries that they should
think of adjustments and sacrifices, if necessary, in their
material living standards in order to gain the benefits, which
are hard to quantify but which may be important, of lower ab-
solute levels of emissions.
It is just as plausible that devel-
oping countries, which are not close to the levels of living in
industrialized countries, would resist a goal formulated in terms
of absolute reductions in emissions.
If developing countries are going to be involved in the
debate over reduction in carbon emissions, a more plausible basis
for comparison is a reduction in emissions relative to what they
would have been, if the country had been followincr a growth path
that was not constrained by emissions reduction
jective that is investigated here.
.
This is the ob-
It is, of course, different
from targets related to the absolute levels of emissions at some
original point in time.
V.
Data base and parameterization
Data requirements of economy wide general equilibrium models
of this nature are quite extensive since a complete, if ag-
gregated, characterization of the economy is required.^
The data
needs can be classified into four broad categories: technological
relationships, behavioral relationships, miscellaneous exogenous
or predetermined variables, and initial conditions.
The estima-
16
tion of these relationships and parameters is described in Blitzer,
et al
(1989)
.
However, since substitution among energy in-
puts in production and consumption has a central role in this
model, the methods used to provide the necessary data will be de-
scribed briefly.
The principal source of primary data on the inter-industry
structure of the Egyptian Economy is a 37 sector transactions
matrix for 1983/84 obtained from CAPMAS.^ The 37 sector matrix is
aggregated into a ten sector classification, adjusted and updated
to represent our base year transactions matrix of 1986/87.
This
transaction matrix provided much of the data for the implementation of the model.
The model is formulated to use one or more technologies to
produce each good or service.
The specific number of alterna-
tives depends on sectoral characteristics. The alternative pro-
duction technologies are divided in two categories.
The first,
encompasses the implicit technologies implied by the transactions
matrix in 1986/87.
The second category of technologies are the
alternatives to the initial technology.
In general, the alterna-
tives allow for substitution between fuels, electricity, labor,
and capital.
The alternative technologies were derived using a
small program which has as inputs:
i)
the initial technology, ii)
the own-price elasticity of energy for the sector; and iii) the
sectoral elasticities of substitution between labor and capital,
between labor and energy, between capital and energy, and between
electricity and fuels.
The model takes the unit demand for fuels
17
as fixed for each technology; but this demand can be met by using
either natural gas or petroleum products.
At the same time,
there are limits placed on the degree to which natural gas and
petroleum products can be substituted for each other.
The methodology used in determining the parameters of the
utility function in the maximand is based on a linear expenditure
system of equations.
The parameters of that function were first
estimated econometrically, and then adjusted for consistency with
the model's base year.
The complete system of consumer demand
functions has (2n-l) independent parameters.
Since these equa-
tions are highly interrelated, a complete systems approach was
used to econometrically estimate the parameters.
The database
for estimating these parameters was constructed by pooling cross-
section family budget data which was available for two time periods,
1974/75 and 1980/81.
Maximum likelihood estimates of the
entire system were derived using the procedure of "seemingly unrelated regression."
As indicated above, the estimates of changes in fuel ef-
ficiency and the capital costs of retrofitting were based on an
examination of the readily available literature.
The estimates
were chosen to reflect cautious optimism as to what is feasible.
However, the authors would not attempt a vigorous defense of any
of their guesses, but, as noted,
represent them only as means of
illustrating the methodology and the general nature of the
results that might be expected.
VI.
The effects on economic performance of restraints
on carbon emissions
The first step is to obtain a "reference" solution to the
18
model, or base case.
This is a solution with only the conven-
tional assumptions with respect to the efficiency with which
fuels are used and without any restrictions on carbon emissions.
The next step is to impose a set of restrictions on carbon emissions, assuming also that these can be perfectly enforced.
Since
the model covers twenty-five years in five year time periods, the
restrictions are also imposed over that period.
The restrictions
imposed are in the form of "umbrella" or "bubble" constraints,
for the economy as a whole, rather than for individual sectors or
even or individual establishments.
Table
1
presents the constraints as percentages of the total
emissions generated in each period in the unconstrained emissions
solution.
The restrictions are stipulated for each future period
as the model, as noted,
is dynamic and extends over 25 years,
capturing the reality of the need for forward-looking policy.
As
will be noticed the emission limits are, in a general sense, in-
creasingly restrictive, over time and in successive solutions.
Table
1
Constraints on Total Carbon Emissions As Percentages
of
Total Emissions in Unconstrained Solution
Gl
G2
G3
G4
G5
1987
'1992
1997
2002
2007
2012
100
100
100
100
100
0.95
0.95
0.90
0.90
0.85
0.90
0.85
0.80
0.80
0.75
0.85
0.70
0.65
0.65
0.60
0.80
0.70
0.65
0.60
0.55
0.70
0.65
0.65
0.55
0.45
19
The constraints are, in a general sense, increasingly restrictive
over time and in successive solutions.
The solutions to the model contains a great deal of detail
with respect to the sectoral patterns of inputs and outputs of
the various sectors in successive periods.
This detail is much
too massive to be presented here and, for the present purposes,
only certain aggregate features of the results are of interest.
Table
2
presents data characterizing in a summary fashion the
results of the solutions.
For the base case and each of the
alternative carbon emission scenarios Table
2
presents an index
of carbon emissions that would be generated during the
I
Table
2
Characteristics of Alternative Solutions
With Different Levels and Patterns
of Carbon Emissions Constraints
Global Carbon
Constraint
Scenarios
Total Carbon
Emissions
Base
Case
Gl
G2
G3
G4
G5
100
94
89
86
85
81
-6.28
-10.64
-14.45
-15.14
-18.88
3.40
2.79
2.48
2.37
2.05
Per Cent Change in
GDP Growth
-3.13
-20.51
-29.34
-32.48
-41.60
Per Cent Change in
Welfare
-0.86
-2.27
-3.43
-4.04
-7.86
Per Cent Change in
Carbon Emissions
Aggregate GDP Growth
Per Cent
3.51
20
first 15 years and the growth rate of gross domestic product (GDP)
over the same period. The percentage change in GDP growth in each
scenario over the base case is also presented
.
Table
2
also
presents the percentage change in consumer welfare over the base
case as measured by the optimized objective function for each of the
cases indicated in Table
1.
To repeat the initial warning, the data are too approximate to
warrant reliance on the results in terms of the absolute levels.
The percentage changes deserve more serious attention, although
these too should be interpreted as more indicative of the range of
possibilities than as predictions.
With this qualification in mind, perhaps the most interesting
features of the results are the non-linear relations between reductions in carbon emissions, on the one hand, and reductions in
measures of economic performance, on the other hand.
The first 6.2 8
per cent reduction in carbon omissions in scenario Gl reduces GDP
growth and the achievable utility only modestly.
In scenario G2
which results in only an additional 4.36 per cent decline in carbon
emissions as compared to Gl, the decline in the growth rate of GDP
is more than 6 times.
Similar non-linearities are shown in the
change in total discounted utility and in the successive scenarios.
It may be recalled that in adjusting to the carbon emissions con-
straints, the model can take advantage of alternative technologies
and sectoral shifts of resources, subject always to resource and
balance of payments constraints.
VII. Experiments with costless improvements in fuel efficiency
The next set of experiments embody the analysis of the potential resulting from completely costless improvements in efficiency
in the use of fuels in several sectors.
Three alternative scenarios
21
Table
3
Alternative Percentage Reductions in Fuel Requirements
for
Natural Gas and Oil
Scenario
Sectors
Petroleum
Electricity
Manufacturing
Construction
Transport
1
Scenario
0.05
0.04
0.05
0.05
0.05
0.05
0.04
0.10
0.05
0.10
2
Scenario
3
0.05
0.04
0.15
0.05
0.15
of this type were tried with the changes shown in Table
3
It might be objected that these are relatively modest improve-
ments in efficiency as compared to the substantial changes that can
be foreseen in the future.
Most discussion of such improvements in-
volve capital costs for new equipment.
Table
4
indicates the consequences of such efficiency improve-
ments when the model adjusts to the set of carbon emissions constraints shown in Table
1.
Part of the adjustments shown in Table
4
result simply from the availabilities of the costless, more fuel efficient technologies.
Those would be used in an optimizing model,
whether or not there were carbon emissions constraints and would
result in increased output and welfare.
It is possible to see this
happening in the background, for example, in scenario Gl, in Table
4.
In this case there is an additional 2.41 per cent reduction in
carbon emissions (which is evident from the percent change in carbon
emissions row), as compared to the same scenario in Table
2,
but a
very small reduction in welfare.
The adjustments are uneven, however, as among the scenarios.
In scenario G3 in Table
4,
for example the reduction in carbon ems-
sions is .63 percent more than in the same scenario in Table
2.
22
Because of the improvements in fuel efficiency, the required
reductions in carbon emissions always have
on economic performance than shown in Table
Table
4
,
less depressing effect
a
2
In scenario Gl in
.
there is an actual improvement in economic performance be-
cause of the provision of new, costless and fuel efficient technologies.
Nonetheless, as the carbon emissions constraints become
more restrictive, economic performance declines substantially,
though never by so much as shown in Table
in consumer welfare in all cases, but,
2
There are reductions
.
again,
less so than if the
This indi-
improvements in energy efficiency were not available.
cates that, although real economic activity is affected less than
before by the carbon emissions constraints, the results are less
satisfying in terms of the optimized utility function.
Table
4
Characteristics of Solutions
With Costless Improvements in Fuel Efficiency of Scenario
in The Adjustment to Carbon Emissions Constraints
Global Carbon
Constraint
Scenarios
Base
Case
Total Carbon
Emissions
100
91
-8.69
Per Cent Change in
Carbon Emissions
Aggregate GDP Growth
Per Cent
Gl
3.51
3.67
Per Cent Change in
GDP Growth
4.59
Per Cent Change in
Welfare
-0.37
Table
5
G2
87
G3
85
1
G5
G4
84
79
-13.15 -15.08
-15.81
-21.18
2.85
2.82
2.31
3.08
-12.34 -18.75
-2.00
-2.71
-19.77
-3.30
-34.13
-7.36
presents the solutions corresponding to the alternative
23
sets of carbon emissions constraints and the costless reductions in
fuel requirements presented in scenario
in Table 3.
2
In this case,
the improvements in fuel efficiency are substantial and GDP growth
rates are always higher than those represented in Table
2
,
and the
decline in the objective function values are also less marked com-
pared to that of Table
2.
Table
5
Characteristics of Solutions
With Costless Improvements in Fuel Efficiency of Scenario
in The Adjustment to Carbon Emissions Constraints
Global Carbon
Constraint
Scenarios
Total Carbon
Emissions
2
Base
Case
Gl
G2
G3
G4
G5
100
90
86
84
83
78
-10.13
-14.24
-16.14
-17.26
-22.37
3.71
3.11
2.88
2.83
2.33
Per Cent Change in
GDP Growth
5.75
-11.40
-18.06
-19.34
-33.70
Per Cent Change in
Welfare
-0.26
-1.87
-2.59
-3.19
-7.17
Per Cent Change in
Carbon Emissions
Aggregate GDP Growth
Per Cent
3.51
With even more substantial, and still costless, improvements in
fuel efficiency, overall performance continues to improve, relative
to the cases without such improvements.
Relatively speaking, the
improvements are much less substantial in the case with the most
restrictive carbon emissions constraints, as compared to the other
cases.
This feature of the results are shown again in Table
6,
which
presents results of the solution with the costless reductions in
fuel requirements shown in scenario
3
in Table 3.
In this case,
as
24
Table
6
Characteristics of Solutions
Improvements
in Fuel Efficiency of Scenario
Costless
With
Emissions Constraints
Adjustment
to
Carbon
The
in
Global Carbon
Constraint
Scenarios
Total Carbon
Emissions
3
Base
Case
Gl
G2
G3
G4
G5
100
88
84
83
81
76
-11.84
-15.53
-17.38
-18.79
-23.67
3.76
3.15
2.90
2.86
2.36
Per Cent Change in
GDP Growth
7.07
-10.31
-17.35
-18.40
-32.79
Per Cent Change in
-0.13
-1.74
-2.45
Per Cent Change in
Carbon Emissions
Aggregate GDP Growth
Per Cent
3.51
I
-3.06
-6.98
Welfare
in the previous two cases,
improvements in fuel efficiency make it
much easier for the model adjust to the carbon emissions constraints.
Yet the degree of satisfaction of consumer utility con-
tinues to fall below the maintenance of economic activity.
And,
in
this case also, the most binding of the carbon emissions constraints
continues to have extremely negative effects on the economy.
Charts
6.
1
and
2
summarize the results shown in Tables 2,4,5 and
The charts help make the point that linear extrapolations would
25
c
I
CO
c
o
CO
CO
N
o
LU
u
O
o
o
c
o
o
c
u
4-1
(/I
\\|
o
CO
4J
c
D
«
«
sz
o
Ld
'
c
o
—
o
C
o
;y
i-
t
<D
o
o
*— I—
a>
S
/
/
c
o
CP
CL
w
\
^"
k'
a
o
N
o
o
o
CM
rM
t\"
tK"
O
O
o
o
o
tf)
n
o
o
fr?
k^
fk"
o
o
o
o
CD
u3
^
IN
o
o
O
o
k"
tK^
a
o
o
i
ci
"w
CJ
suojssjuj^ uoqjDQ
uj UQ!:i::>np»y
%
^
t
(V
0;
^
c
26
CO
i
rM
^
O
E
13
O
CP
^
c
c
o
^
15
O
<-T
-l-i
O
^
D
-M
(J
o
CO
•J
rr
-^--'
r
i
c
o
Kl
O
-)--'
CD
CO
C
o
01
01
E
u
>>.
u
c
c
o
^
1-
o
!t
UJ
U
M
M
V
> o
iD O
4J
-I-'
-M
01
nj
1,'J
<D
S,
^
?
<^
/
S
[^
a
o
^
o
o
IT)
lO
tK^
o
03
O
^
o
o
^
w
o
o
N
^
o
o
Kl
©jDjp/\A uj uc>!:i:>np^y
(J3
d
C
%
It
28
Table 7
Alternative Retrofit Scenarios With Improvements
in Fuel Efficiency
Scenario
Scenario
1
2
Scenario
Scenario
3
Reduction in fuel
coefficient in
manufacturing
0.2
0.3
0.0
0.2
Reduction in fuel
coefficient in
transport
0.0
0.0
0.5
0.5
Payback period
in years in manufacturing
2.5
5.0
4
2.5
Reduction in incremental
capital output/ratio
in transport
10 %
10 %
These scenarios were used in alternative model solutions for
each set of carbon emission constraints. Table 8 presents the
results for Scenario 1 in Table 7 and the alternative emissions constraints in Table 1.
Table 8
Characteristics of Solutions With Improvements
in Fuel Efficiency That Require The Additional Capital
of Scenario 1, Table 7
Global Carbon
Constraint
Scenarios
Total Carbon
Emissions
Base
Case
Gl
G2
G3
G4
G5
85
83
57
91
87
- 9.49
-12.81
-15.02
-16.51
-43.05
3.67
3,03
2.80
2.76
2.30
Per Cent Change in
GDP Growth
4.56
-13.56
-20.11
-21.28
-34.50
Per Cent Change in
Welfare
-0.34
-2.01
-2.75
-3.42
-8.01
100
Per Cent Change in
Carbon Emissions
Aggregate GDP Growth
Per Cent
3.51
30
the changes in overall economic performance are quite modest, except, again,
in the G5 case.
The third scenario in this series involves improvements in fuel
efficiency in the transport sector, achieved through increases in
capital requirements per unit of output.
this experiment are shown in Table 10,
The overall results from
The striking feature of
these results are that the substantial improvement in fuel
Table
9
Characteristics of Solutions With Improvements
in Fuel Efficiency That Require The Additional Capital
of Scenario 2, Table 7
Global Carbon
Constraint
Scenarios
Base
Case
Gl
Total Carbon
Emissions
100
89
G3
G5
G4
':
-11.39
Per Cent Change in
Carbon Emissions
Aggregate GDP Growth
Per Cent
G2
3.51
3.70
Per Cent Change in
GDP Growth
5.27
Per Cent Change in
Welfare
-0.31
86
84
55
82
-14.29 -16.44
-18.16
-45.14
2.79
2.75
2.13
-13.62 -20.51
-21.71
-39.32
-3.46
-8.69
3.03
-2.00
-2.75
efficiency in transport has larger effects on economic growth in all
the cases, as compared to not having such improvements.
However,
the effects on economic welfare, as measured by the objective function, are rather modest, except in the G5 scenario.
Presumably this
reflects the fact that, in the Egyptian economy there is relatively
little use of private automobiles for transport.
Thus the cost of
the use of fuel for transport enters the objective only indirectly,
through the cost of non-transport goods and services.
29
i
The comparisons are becoming more than
a
little awkward.
ever, they may be simplified by reference to Table 2.
How-
It is clear
that exploitation of the option of retrofitting capital in adjusting
to carbon emissions constraints improves the overall economic per-
formance of the economy.
Nonetheless the adjustments to the carbon
emissions constraints generally results in lower real output and
reductions in economic welfare, with the former being larger than
the latter.
The exceptional case is G5, which has the most restric-
tive carbon emissions constraints.
While fitting the general pat-
tern the reductions in carbon emissions, growth and welfare are all
much larger than in previous trials.
It will be recalled that G5 is
the most restrictive set of carbon emissions constraints.
In this
trial, when new, more capital intensive, and less fuel intensive,
methods are forced on the system in order to reduce carbon emissions, there is very little opportunity left for growth.
The rela-
tively large reductions in carbon emissions, larger than required by
the constraints, are the result of much lower levels of economic
performance.
This demonstrates one possibility of which there have
been warnings: that required reductions in carbon emissions, even
with new technologies, may substantially depress economic growth.
Nonetheless the performance of the economy is still better than if
the technological improvements in fuel efficiency had not been
employed.
Scenario
2
in Table 7 is implemented next,
again with the
alternative sets of carbon emissions constraints.
shown in Table
9.
The results are
Although the fuel coefficients are reduced by
fifty per cent more in this scenario as compared to the previous one
32
The effects on carbon emissions and aggregate welfare in the
solutions for all the scenarios outlined in Table
Charts
3
and
4
.
Comparisons with Charts
esting insights.
1
and
2
are shown in
7
provide some inter-
It is clear that the improvements in fuel ef-
ficiency will reduce carbon emissions in both cases.
parison of Charts
2
and
4
However, com-
suggests that forcing changes in fuel use
that are not required by the carbon emissions policy can lead to
larger reductions in consumer welfare than would otherwise take
place.
Table 11
Characteristics of Solutions With Improvements
in Fuel Efficiency That Require The Additional Capital
of Scenario 4, Table 7
Global Carbon
Constraint
Scenarios
Total Carbon
Emissions
Base
Case
100
81
-14.10
Per Cent Change in
Carbon Emissions
Aggregate GDP Growth
Per Cent
Gl
3.51
3.78
Per Cent Change in
GDP Growth
7.69
Per Cent Change in
Welfare
-0.25
IX.
'
G3
G4
G5
77
76
72
-21.19 -22.76
-24.28
-28.11
2.92
2.87
2.42
-10.54 -16.81
-18.23
-31.05
-2.45
-3.03
-6.36
G2
79
3.14
-1.79
Conclusions
It would be misplaced concreteness to claim much in the way of
substantive insight for the results presented here.
These results
are intended to represent the potential of a methodology,
table
31
It is interesting to compare these results with those for the
previous two scenarios, in which improvements in fuel efficiency
were confined to the manufacturing sector.
The fuel efficiency im-
provement in the transport sector was much larger than in manufacHowever, the differential effects were relatively modest,
turing.
except in the case of the most restrictive carbon emissions constraints.
Table 10
Characteristics of Solutions With Improvements
in Fuel Efficiency That Require The Additional Capital
of Scenario 3, Table 7
Global Carbon
Constraint
Scenarios
Total Carbon
Emissions
Base
Case
Gl
100
86
82
-14.12
Per Cent Change in
Carbon Emissions
Aggregate GDP Growth
Per Cent
G2
3.51
3.72
Per Cent Change in
GDP Growth
5.95
Per Cent Change in
Welfare
-0.49
G3
G4
G5
80
79
56
-17.85 -20.27
-21.02
-44.41
2.94
2.87
2.42
-10.77 -16.27
-18.29
-31.20
-3.16
-6.26
3.13
-1.99
-2.63
Finally, in Table 11, results are shown for Scenario
7,
which combines Scenarios
1
and
3
of that Table.
4
of Table
34
Kl
^
o
1-
s:
o
c
t\<
Oi
f/1
S*•
-4-1
s^
M
2
^
+-
4J
in
QC
1
<
^4-
O
n
E
o
-M
5
L-1-"
M
c
o
o
a:
n
c
o
M
iD
CT
£
>^
liJ
o
c
O
e
i>
IN
(5
0°
e,
,^
^
O
^
^?^
?^
L_
^c)^
-
5-C
r
UJ
<J3
oc
a:
^
^
E
/]
/
:v
IS)
I
^
o
o
^
O
O
en
oD
fe^
O
o
fe^
t??
a
o
(??
^"
O
O
t^
o
o
o
o
t^
u)
10
-t
IN
T-
O
o
©J04p/i/\ uj uo!;::>np^y
%
C
•
-
33
rO
D
C
to
tt:
V
g
Pi
^
C3
r
o
o
!^
o
o
c
t
c
o
u
-M
c
'o
i3
4-
1.
m
c
o
O
u
a:
o
-M
^0
3:
O
OT
V
CP
c
G1
«
F
H
LJ
.
O
CS
O
IN
o o
-4-^
Z5
CM
^
13
> c
4J
-M
.y
rvj
i;
-L
c
O
<^
cy
1
^
^
1^
CD
8
o
d
W^
o
o
lo
K
o
o
fr?
a
o
10
10
8
o
6
8
O
10
O
d
8
8
o
d
o
10
8
o
10
8
o
d
•-
IN
SUOJSSJUJ3
uoqjDQ
ui uojijc^np^y
\'\
%
> I/)
36
APPENDIX
Table
A
ParamctcrB and Exogenous Variable
ai
Maximum annual
^t.j.k
Input of good
Sfucl.J.k
Input fuel per unit of production of good J using technology k
Sgas.j.k
Input of natural gas per unit of production of good J using technology k
3pct.j.k
Input of petroleum products per unit of production of good J using technology k
1
rate of depletion of
hydrocarbon resource
per unit of production of good
1
(oil
using technology k
J
bLj.k
Proportion of capital good
technology k
^Lk
Five-year rate of depreciation of capital for production of good
ei
Maximum rate
it
interest rate of foreign debt in year
gl
Minimal post-terminal growth rate
fi.k
ICORi.k
-
1
In the capital required to
of Increase of exports of good
1
produce good
I
1
using
using technology k
between two periods
t
for sector
1
capacity conversion factor for capltol producing good
Incremental capital-output ratio
or natural gas)
for
I
production of good
using technology k
1
using technology k
mi
Demand for labor per unit of production of good using technology k
Demand for labor per unit of agricultural production using technology k
Maximum rate of fall of Imports of good 1 between two jjerlods
qi
Conversion factor for hydrocarbon resource
Sj.k
Maximum share
ll_k
^agr.k
1
1
of natural gas in meeting fuel
(oil
or natural gas)
demand
of producing
good J using
technology k
parameter
consumption good
pi
Elasticity
Yl
Intercept parameter for consumption good
p
Utility
St
Maximum
for
1
discount rate between periods
net foreign borrowing in year
^^
Public consumption of good
Il987
Aggregate imported in 1987
^
Total supply of labor in year
t
^gr.t
Supply of agricultural labor
in
^"t
Population in year
/AXi.t+1
Discoveries of resource
In year
1
year
t
t
t
t
1
(oil
or natural gas) between year
Tt
Other foreign exchange transfers
FPi-
Foreign firms' profit rcmittaLnccs in year
Wt
Workers' remittances
in
year
Pu
world price of expons
at
good
Pl.t
1
world price of Impons
at
good
in
year
t
in
1
l
year
m year
t
t
t
t
t
and year t+1
55
The substantive results which do exist originate in the com-
parison of the alternative scenarios and cases.
The demonstration
of nonlinearity in the economic impact of carbon emission con-
straints appears to the authors to be a robust outcome.
There are
no features of the model that would necessitate such an effect.
The subtleties in the adjustments to changes in efficiency and
emission constraints and the differences in the adjustments depending on the strictness of the constraints appears to be additional
robust results.
They warn against simplistic diagnoses and policy
prescriptions
While complex by comparison to many economic models, this model
is still a relatively simple depiction of an economy.
Nonetheless
its complexity is sufficient to provide an important warning: carbon
emissions constraints, depending on their severity, can have far-
reaching and profound economic effects.
would be playing Polyanna.
To convey any other message
.lb
Table
5
Endogenous Variables
^
Net foreign borrowing In year
Ci,t
Private
Dt
Foreign debt In year
Ei.t
Exports of good
I^t
consumption
good
t
1
In year
t
Investment demand for good
1
In year
t
Il.J.k.t
Demand
1
by sector J, technology
Ki.k.t
Installed capacity In year
^^.k,t
New
Mi,t
Imports of good
Pl.t
Shadow price
Rl,t
Reserves of hydrocarbon
U(Ct)
Utility of
W
Total discounted
Xii
Gross domestic output
of
Xi.k.t
Gross output of good
produced using technology k.ln year
Zix
Intermediate deliveries of good
1
of
t
In year
for Investment
t
good
t
produce good
to
capacity to produce good
of
1
in year
1
i
k. In
year
t
using technology k
using technology
k,lirst available in
year
t
"
good
I
in year
1
(oil
t
or natural gas) in year
per capita consumption in year
utility:
1.
the
t
t
maxlmand
good
1
in year t
1
In year
t
t
Vit
Total amount of carbon generated by the use of
particular fuel, i, in period, t
Vijt
Total amount of carbon generated by the use of a
particular fuel, i, in sector j, in period, t
Vii^jt
Amount of carbon generated by the use of
fuel, i, using technology k, in sector
j
V^ct
Vikjt
,
t
a
a
in period, t
Amount of carbon generated by the use of
fuel, i, in consumption in period, t
a
particular
Quantity of carbon emission per unit use of
particular fuel, i, using technology k, in sector
in period, t
j
,
v^jt
Quantity of carbon emission per unit use of a fuel,
in consumption in period, t
i,
.),
Vt
Maximum amount of carbon that may be generated in
period, t
Vit
^ikjt
Maximum amount of carbon that may be generated, by
sector j, in period, t
Maximum amount of carbon that may be generated, by the
use of a particular fuel i, using technology k, in
sector j, in period, t
40
R^.ui
=
Ri.t
+ ARi.ui
-
2.5{Xcui + Xu)qi
1
Dui = Dt +
2.5(Bt^i +
10/
BJ
(16)
Invrsfmpnt
Drmrnd
J
Ii.j.tt
=
bij.k
kt
(17)
ICORj.k
AKj.k.t+1
(18)
X
^1-1987
2- ^k.2017 ^
k
^
1
1987
(19)
(1 +^l)
Xk '^k.2012
(20)
Carbon Emissions
(21)
"ijt
-I
V
ikjt
k
V.
It
-I
V.
(22)
.
ijt
J
V.
ict
— V.
c.
let it
\ijt-^kijt\jt
V
ijkt
V.
It
[
< V
*
(23)
(24)
(25)
ijkt
< V.
(26)
It
(V,, + V.^^) ^ V^
(27)
59
MODEL
Accounting Idrntltlrs
Xi.t
+
Mi.t
=
+
Zi.t
+
Ci.t
+
G,.t
Ii.t
+
E,.t
(1)
ir
k
J
Z Pu
Ei.t
(2)
(3)
Mu
X Pu
+ Wt + Tt + Bt =
+
it
Dt + FP^
(4)
Technology and Production Constraints
agas.
J.
k +
agas.J,
XS
J.
k =
afucl.
J.
k
k - ^.k^fucl-J. k
^1.
k X(.
^
k.t
(5)
(gj
Lt
kv
1
X,
Spct.
(7]
^agr.k Xagr.k.t
^
-U^r.t
k
(8)
^.k.t ^ Ki.k.t
qi Xi,
t
^
ai R(.
(9)
t
[10)
Balanrr of
Privrr^P^^c p nd •^Ynr'e rnr.ytr^inf^
Bt < Bt
Mi.t
Ett
Dv-narrJc
> (1-mJ
(111
Mi.t-i
(12)
^ (l+ei)Eu-i
!.!nk.->^T>s
K4.k.[-1
=
K4.i<_i(l-di.k)
+
fi.k
AKi.i^c
(1'
.
.
.
.
.
'^^
NOTES
The focus of the paper should not be interpreted as a judg1
ment on the dangers, actual or incipient, of a "greenhouse" effect, due to carbon emissions, on which no position is taken
It is, rather, an exploration of the consequences of some
here.
of the policies proposed in anticipation of the problem.
See R. Ayres, (1989)
2
3Several attempts have been made to quantify these effects including D. Jorgenson and P. J. Wilcoxen, (1989) A. Manne and R.
Richels (1989), W. Nordhaus, (1989), and C. Blitzer, et
,
al, (1989)
4See H. Leibenstein,
.
(1987)
(1966)
It should be recalled that the purpose in presenting the
model is primarily methodological. The omission of coal as a
primary energy source would, of course, be quite wrong for most
countries, although correct in the case of Egypt.
6Blitzer, et al
(1989)
7
See Blitzer, et al
(1989)
8Central Agency for Public Mobilization and Statistics (CAPMAS)
,
5
.
.
41
Objective Function
(28)
t
(c.
l.t
u(c,) -
I ^ii.
-
7-
(29)
43
BIBLIOGRAPHY
"Energy Conservation in the Industrial Sector,"
International Institute of Applied Systems Analysis,
Laxenburg, Austria, 1989.
Ayres, Robert,
Eckaus, Richard S., Lahiri, Supriya and
Blitzer, Charles R.
Meeraus, Alex, "An Economy Wide Energy Policy Model for
Egypt," Center for Energy Policy Research, MIT, Cambridge
July, 1989.
Mass.
,
,
Jorgenson, Dale and P.J. Wilcoxen, "Environmental Regulation and
U.S.
Economic Growth," presented at the MIT Workshop on Energy
and Environmental Modeling, July 31-Aug. 1, 1989
Lave,
Lester B.
"The Greenhouse Effect: What Government Actions
Are Needed?" Journal of Policy Analysis and Management
7,3, 1988, 460-470^
,
,
Leibenstein, H.
"Allocative Efficiency vs X-ef f iciency
ican Economic Review 56, June, 1966, 392-415.
,
,
"
Amer
,
Inside the Firm: The Inefficiencies of Hierarchy
Press, Cambridge, Mass., 1987.
,
Harvard
Manne, Alan and Richels, Richard G. "C02 Emission Limits: An
Economic Analysis for the USA," presented at the MIT Work
shop on Energy and Environmental Modeling, July 31-Aug. 1,
1989.
Nordhaus, W.
"To Slow or Not to Slow: The Economics of the
Greenhouse Effect," 1990, (mimeo)
"Economic Growth vs. Climate Change: Survey of Policies to
Cope with the Threat of Climate Change," 1990, (mimeo).
,
1181
025
U.
7 '10
16
Date Due
NO 2
JIJ.2 9
7 '92
1994'
DUE FEB
- lv98
Lib-26-67
MIT LIBRARIES
3
IDflD
D0b5a3T4
T
Download